AI Stocks Lose $6 Trillion. Pension Funds Face Hidden Exposure Risks.
Investment managers are reassessing risks as new research estimates that U.S. public pension funds hold up to $706.6 billion in AI-related assets.

A selloff that wiped more than $6 trillion from global artificial intelligence stocks has prompted investment managers to reassess how much of their portfolios depends on the technology, including holdings that may not appear directly connected to AI.
The decline between Sept. 30 and Nov. 6, calculated by MSCI, has renewed questions about whether investors can adequately diversify their portfolios as AI-related companies account for a growing share of global stock markets.
The concerns extend beyond technology shares to corporate bonds, businesses that supply data centers and public pension funds holding retirement savings for teachers, police officers and other government workers.
According to Reuters, at least $25.3 trillion is invested in major global companies with direct exposure to AI, based on the MSCI Global Intelligence Index. The figure represents investments in those companies, not money lost during the selloff.
The wider exposure is harder to measure because AI affects businesses through their customers, suppliers, operating costs and investment plans.
Andy Goldberg, chief investment strategist at Nomura Asset Management, told Reuters that AI has become a highly interconnected investment theme, making it difficult for investors to determine how much risk they are carrying.
He pointed to companies outside the technology industry that are increasingly incorporating AI into their operations, including retailers and transportation businesses.
The question has become more pressing following recent stock market declines tied to AI spending concerns and warnings about the technology's potential implications for U.S. financial stability.
MSCI has developed its AI Value Chain Indexes to help investors identify exposure across 10 parts of the industry's development, from semiconductor manufacturing and data centers to software applications.
Its research found that U.S. companies represented almost 80% of the broad AI Value Chain Index at the end of October, compared with approximately 64% of MSCI's wider global equity index.
Four industry groups, semiconductors, software and services, media and entertainment, and technology hardware, accounted for more than 70% of the AI index's weight. Those same industries represented about 35% of the broader market index.
The concentration means investors who own several funds or market indexes may still be exposed to many of the same AI-related companies.
MSCI also found that its AI-related physical infrastructure index fell 22% between Sept. 30 and Nov. 6, while digital infrastructure declined 10% and AI applications dropped 6%.
Industries with lower correlations to AI performed better during the same period, although MSCI cautioned that the relatively short timeframe was insufficient to establish whether the pattern would continue.
Some major investment managers are already examining their portfolios more closely.
PGIM, the investment management business of Prudential Financial, has begun evaluating nearly $1 trillion in fixed-income investments using two AI-related screening methods, according to Reuters.
One examines how AI could affect individual companies' business models and creditworthiness. The other assesses how investments could respond to developments such as a slowdown in data center spending or broader market disruption.
Brian Barnhurst, PGIM's head of global credit research, said direct investments in semiconductor companies or data centers were easier to identify than indirect risks affecting other industries.
Those indirect exposures could include businesses whose growth depends on AI infrastructure projects or companies vulnerable to changes in technology spending.
Public pension funds are another area of concern. A November report from the Equable Institute estimates that U.S. state and local pension systems held between $578.1 billion and $706.6 billion in AI-related assets as of June 30.
The range represents approximately 9% to 11% of the $6.42 trillion held by those pension systems, according to Equable's announcement.
The research also found that AI-related companies accounted for 35.1% of reported public equity holdings, highlighting how heavily some retirement portfolios depend on technology businesses.
The investments were concentrated in a relatively small number of companies.
According to Equable, 15 companies accounted for approximately 85% of the AI-related holdings it examined.
Seven companies, Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla, accounted for 63.3% of those holdings.
The concentration reflects the importance of major technology companies in stock indexes, including those followed by pension funds and other institutional investors.
Equable's estimates do not mean that pension funds have lost the amounts identified in its report. They measure the value of holdings classified as AI-related, some of which may also benefit from future growth in the technology.
However, the findings illustrate how a downturn affecting the same group of companies could have consequences beyond investors who deliberately purchase AI-focused shares.
The risks have become increasingly relevant as Wall Street questions whether the billions being committed to new computing infrastructure will generate sufficient returns.
Recent technology stock declines have also highlighted concerns about valuations and the pace of AI investment.
Jay Jacobs, U.S. head of equity ETFs at BlackRock, told Reuters that the growing dominance of AI-related companies within major indexes was making diversification more difficult than some investors recognize.
That concentration has been evident during recent selloffs involving Amazon, Intel and other technology companies.
Some investors nevertheless see opportunities beyond the companies most closely associated with the AI boom.
Reuters cited investment strategists who believe that businesses benefiting from AI adoption could eventually offer attractive returns even if the suppliers of chips and computing infrastructure experience greater volatility.
For institutional investors, the immediate challenge is determining where their exposure lies and whether apparently separate investments are vulnerable to the same changes in AI spending and market sentiment.
© Copyright IBTimes 2026. All rights reserved.

